Democratization and the depoliticization of the banking sector: Are all banks affected equally?

Abstract

This paper shows that the influence of the dramatic change in political ties of the Indonesian banking sector following democratization varies widely by bank ownership type. Panel estimates using several unique data-sets show that the decrease in political ties over democratization positively influences the performance of government-owned banks while negatively influencing the performance for privately held banks. Results remain robust to a variety of alternative hypotheses and sensitivity tests. This study provides evidence not only of the differential impact of depoliticization, but also suggests that the relationship between political connections and performance may be different based on bank ownership.

Keywords:

• democratization
• governance
• political connections

JEL classifications:

• G21
• N25
• N45
• P16

View correction statement:

Erratum

1. Introduction

From the first recorded Lulin rebellions in 17 AD to the Arab Spring, public frustration with elite capture of important political and economic institutions has historically played a central role in the transition to democracy. Democratization is seen as a salve; if it successfully ushers in institutional changes that equalize the right to allocate political and economic resources (Carnes and Lupu 2015; Cheibub, Gandhi, and Vreeland 2010; Freeman and Quinn 2012; Haggard and Kaufman 2012; Hollyer, Rosendorff, and Vreeland 2011; Limongi and Przeworski 1993; O’Donnell, Schmitter, and Whitehead 1986), then it should be the case that political influence and rent-seeking in economically important sectors should dissipate. As the banking sectors of authoritarian regimes are typically controlled by politicians or their cronies (Chekir and Diwan 2014; Haggard and MacIntyre 2001; Morck, Yavuz, and Yeung 2011), the transition to democratization should thus correspond with a loosening of the grip on financial resources by the elite (Dryzek 2009).

While prior work contends that democratization should lead to changes in the political ties in sectors like banking (Acemoglu and Robinson 2006; Geddes 1999; Geddes, Wright, and Frantz 2014), it is unclear if all banks are affected equally by this change. This is important to understand for several reasons. From a policy perspective, understanding how changes in political connectedness affect banks differently provides a needed nuanced perspective into banking sector reform. If the changes accompanying democratization affect banks differently, a “one size fits all” approach may not be suitable. The many countries worldwide implementing holistic banking sector reform post-Great recession, for instance, would be well advised to gain a better understanding of whether different types of banks are impacted differentially by this substantive change.

Furthermore, analyzing how these changes affect banks differently provides fundamental insight into the nature of political connectedness that the prior literature has yet to establish. Theory and subsequent empirical results on the impact of political influence in banking are mixed; while it is logical from the preferential access view that political connectedness may provide special treatment that would enhance performance (Faccio 2006; Faccio, Masulis, and McConnell 2006; Fisman 2001; Khwaja and Mian 2005), if political connectedness subjects a bank to being a vehicle with which politicians pursue political rather than profit-motivated goals, then the political objectives view suggests the connectedness of banks may in fact be detrimental to banks’ financial performance (Baum, Caglayan, and Talavera 2010; Dinc 2005; Shleifer and Vishny 1994).

Here, we posit that the role of political connections differ based on whether a bank is publicly owned by the government or privately held. Namely, government-owned banks may be influenced by the inefficient goals of politicians as these banks benefit from a government safety net, and are controlled by the government. Privately held banks, conversely, use political influence to their benefit by seeking out connections that will provide preferential access in order to improve performance. Thus, examining whether depoliticization differentially impacts banks provides a key insight into the very nature of political connections; that in fact while the literature aims to categorize connections as “bad” or “good” for an organization, they may in fact serve difference purposes for different banks.

This paper explores the varying impact on banks of the depoliticization following democratization using the Indonesian experience of 1999. Democratization was ushered in largely as a solution to the corrupt distribution of valuable resources to the elite that had become institutionalized during the Suharto regime (Webber 2006). Over the 32-year regime, this institutionalization of favoritism leads to the barrier between public and private sectors became almost non-existent. Thus, one would expect if the effects of democratization to be true, that political connectedness in important sectors like banking to change.

Using a unique data-set that measures the political connectedness of bank board members for the population of Indonesian banks from 1993 to 2008, the results provide evidence of the widespread depoliticization of the banking sector in Indonesia following the democratic transition. Furthermore, panel estimates show that this change in political connectedness was associated with different changes to performance between privately held and government-owned banks. On the one hand, government-owned banks benefitted from the depoliticization following democratization. On the other hand, privately owned banks performance suffered as political ties changed. Results remain robust to a variety of considerations and sensitivity tests.

These results provide evidence that the changing political structure that democratization prompted had a very different affect for different types of banks. Furthermore, results provide support for the idea that the role of political ties in terms of performance varies for different banks. Thus, while the literature attempts to provide evidence for either the preferential access or political objectives view on the impact of political connectedness, the reality may be that both exist, yet just for different types of banks.

While this paper examines one large emerging market in detail, many other developing and emerging economies worldwide have experienced similar regime changes. By some accounts, over 85 authoritarian regimes have transitioned to democracies such as Indonesia between 1974 and 1999. Even more recently, countries such as Tunisia, have experienced uprisings and subsequent transitions from autocracy to democracy. Thus, understanding how changes accompanying transition alter political connectedness and financial outcomes in the banking sector is not a phenomenon reserved for one country alone. The experience of Indonesia, therefore, can be used to better understand or at least serve as a benchmark to compare to other countries experiencing such transitions.

The rest of the paper is organized as follows. In Section 2, we discuss the Indonesian context. Section 3 then discusses and describes the unique data used to analyze the main questions. Section 4 presents the results of the panel analysis. Section 5 goes in-depth into discussing robustness and sensitivity tests. Section 6 discusses remaining limitations and concludes.

2. Democratization and depoliticization in Indonesian banking

The call for democratization in Indonesia began resonating towards the end of the Asian financial crisis of 1997. While Indonesia had been experiencing steady economic growth for many years, the crisis leads to a sudden, major deterioration of the Indonesian economy. Within one year, GDP dropped by 13.5%, the Indonesian rupiah lost half of its value, and basic needs such as milk for children became inaccessible (Furman et al. 1998). The crisis in Indonesia is viewed as particularly dramatic because economic growth prior to the crisis was steady, providing little foresight that the economy was vulnerable (Haggard and MacIntyre 2001).

It was perhaps this steady economic growth that also contributed to the public’s tolerance of the authoritarian rule and transgressions of President Suharto from 1967 to 1998. During his rule, President Suharto became notorious for siphoning money and other resources from enterprises for himself and his group of cronies. Companies with connections to him, for instance, enjoyed economic advantages such as protection from competition, access to financing, and tax benefits (Cull and Xu 2003; Cull et al. 2015; Dieleman and Sachs 2008; Fisman 2001; Johnson and Mitton 2003; Wu et al. 2012). The connection between banking and politics was so intimate, for instance, that several of the largest banks in Indonesia were at least partially owned by Suharto relatives (Poczter 2015).

Yet once the economy deteriorated, the elite control over economic resources became unacceptable by the populace (Crouch 2000; Webber 2006). And so, by 1998, widespread rioting and a general atmosphere of political chaos lead to the resignation of Suharto and a call for the process of democratization to begin. After Suharto’s stepping down, much thought was devoted to fundamentally changing the political system and ending corruption in all branches of the state (Webber 2006). This lead to both explicit reforms intended to end the system of political favoritism and more implicit changes ending the culture of blind acceptance of the President and his network’s authority (Crouch 2000; Sharma 2001).

While the details of the new democracy would not be completely settled until 2004, the most significant changes were in place upon the first democratic elections in 1999. For the first time, electoral competition was not restricted in terms of the number of parties and there was more than one single candidate for President (Horowitz 2013). After 1999, the managed political competition under the authoritarian regime transitioned to a competitive multiparty democracy with dozens of political parties competing for power.

Over the financial crisis and democratization period, the banking sector also experienced a great deal of turmoil. In the years prior, just as the country experienced a period of steady economic growth, so too did the banking sector. From 1992 to 1996, for instance, outstanding bank credit increased by over 20% per year (Sharma 2001). The growth of the banking sector was largely the result of widespread liberalization reform during the 1990s, intended at least in part to reduce the dominance of state-owned banks by promoting the entry of privately held banks (Enoch et al. 2001). State-owned banks had originated in Indonesia as part of industrial policy following Indonesia’s independence from the Dutch, and were the main source of financing in Indonesia thereafter. The government exerted a great deal of influence on the state-owned banks; managers were notoriously subjected to the political pressures of the Suharto network and high-ranking officials (Sharma 2001). As a result of the liberalization, the banking sector experienced a rapid growth of privately held banks to compete with the state-owned banks.

While questions did arise as to the long-term effects of the easing of credit that deregulation provided ex-poste (Michael 1995), the general consensus of the strength of the banking sector at the time, however, was a positive one (Sharma 2001). By 1996, there were 278 commercial banks, 43 government-owned (including regional development banks), and 235 privately owned (including domestic private, joint venture, and foreign banks). Table 1 provides a sense of the general state of the banking sector based on ownership type in 1996. Here, we see that lending by the privately held banks was slightly larger than that of government-owned banks on an absolute basis, yet on a per-bank basis, government-owned (and specifically state-owned) banks were significantly larger in terms of lending (columns 1 and 2). It is also evident that there was a large variation in performance among the different types of banks. While state-owned banks had the highest net interest margin and solvency, foreign-owned banks were the highest performing in terms of return on assets and return on equity.

Table 1. Bank-level descriptive statistics for the population of Indonesian banks prior to democratization (1996).

	(1)	(2)	(3)	(4)	(5)	(6)	(7)
	Number of all banks 1996	Number of banks 1999	Lending	Return on equity	Return on assets	Net Interest Margin	Solvency
Domestic private	183	82	168,256	10.154	1.51	3.565	0.057
Foreign	14	14	12,263	31.323	4.191	4.852	0.052
Joint venture	38	25	15,307	20.492	2.436	3.751	0.031
Regional banks	30	25	7748	27.533	2.403	4.562	0.032
State banks	13	5	158,258	15.451	0.912	5.156	0.061
Total	278	151	361,832	20.892	2.105	4.827	0.052

Notes: Column 2 reports total lending by each group in billions rupiah. Columns 3–6 report the averages within each group.

The financial crisis was severely debilitating for the banking sector; large-scale bank runs helped catalyze the closure of a significant number of banks. Table 1 shows that over half of the pre-existing domestic private banks no longer existed by 2001. Similarly, the number of government-owned banks decreased by approximately 30% as the result of either consolidation or closure. As a result of the deterioration of the banks’ condition, many government-owned and privately held banks came under supervision of the Indonesian Bank Restructuring Agency (IBRA), a government program established in order to implement bank recapitalization, recover bank assets, and recover government funds disbursed to the banking sector. The intention of the recapitalization program was to contain the crisis in the banking sector, restore lending, and restructure banks in the spirit of better governance (Poczter 2015; Sato 2005). The banking sector was perhaps not considered fully “recovered” from the crisis largely until late 2004, when IBRA was finally closed (Chua and O’Rourke 2002).

3. Data

We use several data-sets in order to measure the change in political connectedness over the democratization period and how this differentially affected government versus privately held banks. Our main data-set contains annual information on the population of Indonesian banks from 1993 to 2008 at the bank level. Obtained by Bank Indonesia (Indonesia’s central bank), this data-set contains the variables we use to identify connectedness (the names of bank board members), measure bank lending and performance, and identifies the ownership structure of each bank (government-owned, privately held, etc.). This data-set contains financial and descriptive annual information for every bank in Indonesia. Each bank submits on set of financial statements and information at the bank level such that this information includes the aggregate of all bank branches.

This data-set provides detailed information of the governance of banks as well. This includes the names of those on the board of directors, supervisory board, board of commissioners, and shariah boards (for banks that provide shariah financing) at the corporate level. While Indonesian banks are required to have a two-tiered board consisting of the board of directors and an oversight-related board of commissioners at the corporate level, many banks also choose to have additional boards such as audit boards, and compliance boards for supplementary oversight and governance.

We also use another data-set compiled from the curricula vitae of members of those member of the political elite of Indonesia from 1993 to 2009. This unique information was hand-collected by the author and combines two underlying data-sets. The first, The Encyclopedia of Prominent Indonesians, is a comprehensive database containing profiles of formal and informal leaders, politicians, businesspeople, experts and other individuals with political influence, updated on a continual basis. This data-set was originally compiled by Tokoh Indonesia online, an organization consisting of journalists and former journalists intending to increase transparency in Indonesia by publicly cataloguing those with political influence.¹ The other data-set used to compile the CVs of the political elite was information obtained for all members of the bicameral Indonesia legislature elected in 2001 and 2005 (the House of Representatives (DPR) and House of Regional Representatives (DPD)). All members of the DPR and DPD are asked to provide a CV to the government once elected to office, and a member of administrative body of the DPD privately provided this information to the author. These additional CVs were used to both verify and supplement the Encyclopedia of Prominent Indonesians.

All in all, the combined elite data-set includes CVs for over 1600 political elites active in the pre- and post-Suharto period. The CVs cover career information such as jobs and political positions held, demographic information such as place of birth and religion, and additional information regarding education, and political party affiliation. For further detail on this data-set, see Poczter and Pepinsky (2016) and Appendix 1.

The two main data-sets were then constructed by creating an algorithm to match the names of the political elites to the names of all of the board members of each Indonesian bank for each year from 1993 to 2008.² Thus for each bank-year, it is possible to observe whether a member of the Indonesian political elite was also a board member. Again, we consider a board member to be a member of all of the several types of bank boards. In this regard, the denominator for the metric of proportion of bank boards that are elites, for instance, is a sum over all the members of all boards of the bank.

3.1. Summary statistics

3.1.1. Political connectedness of banks by ownership type

Table 2 provides information about the number of political connections on all bank boards by ownership type for the set of surviving banks in order to examine the change in connectedness over democratization. Here, we show the number of unique\bank-elite connections over all years (column 1), that existed in the pre-democratization years (column 2), in the pre-democratization years only (column 3), post-democratization years only (column 4), and those connections that were maintained over both time periods (column 5). Over this period, 35% of banks had at least one political elite on the bank board, representing 202 unique bank-board member connections. The vast majority (75%) of these political connections were to privately held banks that were domestically owned. Altogether, the privately held banks had many more political connections in number. Considering, however, that privately held banks outnumbered government banks by about four privately held banks to each government-owned bank, the average number of connections is about equal.

Table 2. The number of political connections of Indonesian bank boards over democratization 1993–2008.

	(1)	(2)	(3)	(4)	(5)	(6)	(7)
	All Years	Total pre-democratization	Pre-democratization only	Post-democratization only	Both periods	Exit rate	Decrease in connections over period
Domestic private	152	134	72	18	62	0.53	0.41
Joint venture	14	13	4	1	9	0.30	0.23
Regional bank	19	17	10	2	7	0.59	0.47
State bank	17	15	12	2	3	0.80	0.67
Total	202	179	98	23	81	0.55	0.42

Notes: Column 1 indicates the total number of unique bank-elite connections over all of the years in the sample. Column 2 shows the total number of ties that existed pre-democratization (1993–1999). Column 3 shows the number of unique bank-elite connections that only existed in the pre-democratization years. Column 4 shows the number of unique bank-elite connections that existed in only the post-democratization years (2000–2008). Column 5 shows the number of unique bank-elite pre-democratization connections that existed prior to democratization and remained over the transition. Column 6 shows the exit rate of political ties (column 3 divided by column 2). Column 7 shows % decrease in connections as calculated by summing columns (3) and (4) and dividing by the number of pre-democratization connections (column 2 plus column 4), thus taking into account new ties emerging post-democratization. Foreign banks are excluded as they do not exhibit political connections. This table includes both surviving and closed banks.

It is evident here that over the democratization period, there was a decrease in the number of connections for all types of banks. Columns 3 and 4 of Table 2 begin expressing this decrease by showing that a large portion of the unique bank level ties existed only in the pre-democratization period, and very few new ties were made in the post-democratization period. Column 6 shows the exit rate of political connections that existed in the pre-democratization period. Here, we see that the exit rate varied from 30% for joint venture banks to 80% for state-owned banks. Namely, 30% of the ties that existed for joint venture banks in the pre-democratization period no longer existed post-democratization. Column 7 shows the decrease in political connectedness over this period, considering new post-democratization ties as well. This shows that for the entire sector, the political connectedness decreased by about 42% over the democratization period.

We also examine the political connectedness of closed banks in order to get a sense of whether the depoliticization we observe in Table 2 is driven by bank closure. Table 3 indicates that approximately 30% of the elites that exited were bank members of closed banks. As approximately 30% of the number of banks exited, the number of political connections of the closed banks is proportionate to the number of banks exiting. So it is not the case that a disproportionate proportion of connections exited.

Table 3. Number of political connections by bank ownership type for closed banks.

	(1)	(2)	(3)	(4)	(5)
	Total number pre-democratization ties (closed and surviving)	Pre-democratization only (closed and surviving)	Political connections to closed banks	% Total pre-democratization ties from closed banks	% Ties of pre-democratization only
Domestic private	134	72	20	0.15	0.27
Joint venture	13	4	0	0	0
Regional bank	17	10	3	0.18	0.30
State bank	15	12	4	0.27	0.33
Total	179	98	27	0.15	0.28

Notes: Column 1 indicates the total number of unique bank-elite connections that existed in the pre-democratization period. Column 2 shows the total number of ties that existed pre-democratization only (i.e. no longer existed after democratization). Column 3 provides the number of unique elite connections to closed banks. Column 4 and 5 consider column 3 as a proportion of column 1 and column 2, respectively. Foreign banks are excluded as they do not exhibit political connections.

We examine an additional metric to establish the depoliticization over the democratization period in order to take into consideration board size. Table 4, therefore, looks at the average proportion of elites on each bank board by ownership type as well, separating between the pre-democratization and post-democratization years. Column 1 indicates that over all years, approximately 30% of the board members in the banking sector were members of the political elite. This average was largest for state-owned banks, with 35%, and lowest for joint venture banks, with only 12% of board members being members of the elite.

Table 4. Proportion of Indonesian bank boards over democratization 1993–2008.

	(1)	(2)	(3)
	All years	Average proportion of political elites pre-democratization	Average proportion of political elites post-democratization
Domestic private	0.25	0.33	0.21*
Joint venture	0.12	0.15	0.11*
Regional bank	0.28	0.39	0.26*
State bank	0.35	0.41	0.33*
Total	0.33	0.38	0.25*

Notes: The proportions of Table 4 are calculated in the following way. The proportion of political elites on each bank board for each year is calculated. Then, the average proportion for each year, the average proportion for each type of bank based on ownership type per year is calculated. Then we calculate the proportions over each time period.

* t-test of the differences in the means presented in columns 2 and 3 is significant. This includes closed banks.

From the banking sector level perspective, this proportion decreased over democratization. Column 2 shows that while prior to democratization, the average proportion of political elites on bank boards was 38%; column 3 indicates this proportion decreased to 25%. Further, we see here as well that a great deal of variation existed among banks based on ownership type in the pre- and post-democratization era as well. While domestic private banks decreased from 33 to 21%, state banks decreased from 41 to 33% over this period.

Table 5 examines the differences in the pre-democratization proportions for surviving versus closed banks.³ This table shows that the banks that closed did not have a statistically larger proportion of elites on their bank boards relative to surviving banks except for domestic private banks. This suggests that changes to political connectedness using this metric for surviving banks are not affected significantly by attrition.

Table 5. Differences in the proportion of political connectedness by bank ownership type between remaining and closed banks.

	(1)	(2)
	Pre-democratization surviving banks	Pre-democratization closed banks
Domestic private	0.33	0.41*
Joint venture	0.15	0.14
Regional bank	0.39	0.38
State bank	0.41	0.38
Average across all banks	0.38	0.26

Notes: The proportions of Table 5 are calculated in the following way. The proportion of political elites on each bank board for each year is calculated. Then, the average proportion for each year, the average proportion for each type of bank based on ownership type per year is calculated. Then we calculate the proportions over each time period.

* t-test of the differences in the means presented in columns 1 and 2 is significant.

4. Regression results

To get a sense of how the decrease in political connectedness differentially affected government-owned versus private banks, we use a panel fixed effects regression split-sample approach.⁴ In one sample, we look at the relationship between the proportion of political elites on bank boards and performance metrics over democratization using only government-owned banks. We include both the state-owned and regional banks for several reasons. First, Table 2 provides evidence that the political connectedness of regional banks is not insignificant. This is consistent with increasing evidence that regional level political influence is meaningful (Li and Zhou 2005; Menozzi, Gutiérrez Urtiaga, and Vannoni 2012; Wang, Wong, and Xia 2008). Second, the regional banks closely resemble the state-owned banks in that they are owned and controlled by the government albeit on the regional and not federal level. Further, even though these banks are much smaller than the state-owned banks, they also operate under a pro-social purpose, aiming to ensure access to banking services for people across the Indonesian archipelago (Sharma 2001).

In this analysis, our main variable of interest is the political connectedness of banks, measured as the proportion of political elites of all bank board members on the bank-year level. We interact this variable with democratization indicator equal to one for each year post-democratization (1999) and zero for each year prior to democratization. The main parameter of interest, therefore, measures the differential impact of political connectedness over democratization for each type of bank.

In addition to this main interaction effect, each regression includes bank fixed effects controlling for observable and unobservable time-invariant bank heterogeneity and a year indicator, controlling for changes each year that affect all banks. The regression also controls for time-varying bank level variables including total board size, the competitiveness of the bank’s industry (a geography-based measure of industry competitiveness using on the share of deposits of banks in the industry), operating leverage (fixed assets/total assets), liquidity (liquid assets/liquid liabilities), and bank size (the natural logarithm of total assets).

It is also worth discussing the importance of including board size in each regression. Including board size is important in order to clarify whether the depoliticization is just a result of decreases in board size. Thus, by controlling for board size, our results show that the changing relationship between political connectedness and bank financial outcomes is not driven by decreases in board size.

Tables 6 and 7 present the main results. We examine the differential relationship between the changes in political connectedness of banks over democratization using different bank-level financial outcomes (lending, return on equity, return on assets, net interest margin and solvency) for government-owned and privately held banks, respectively. Table 6 indicates that political connectedness is associated with an increase in lending and decrease in performance for all performance metrics for government-owned banks. This negative relationship, however, decreases over the democratization period. In other words, the negative influence of political connectedness on the performance of government-owned banks decreases over the democratization. This relationship varies in magnitude across performance metrics. For instance, the return on equity increases by approximately 10% as a result of the average change in political connectedness for government-owned banks over this period, while that for net interest margin decreases by about 5%.⁶

Table 6. The relationship between bank political connections and outcomes for government-owned banks 1993–2008.

	(1)	(2)	(3)	(4)	(5)
Dependent variable	Lending	Return on equity	Return on assets	Net interest margin	Solvency
Proportion of political elites on bank board * Post	0.196***	0.018*	0.001*	0.017**	0.020**
	(0.022)	(0.020)	(0.000)	(0.006)	(0.003)
Proportion of political elites on bank board	0.188***	−0.014***	−0.012*	−0.013**	−0.018*
	(0.004)	(0.045)	(0.005)	(0.002)	(0.005)
Post	0.120*	0.201*	0.100*	0.0839**	0.102*
	(0.010)	(0.001)	(0.005)	(0.021)	(0.009)
Constant	8.615**	2.555*	0.669*	0.695***	5.621*
	(2.005)	(0.909)	(0.209)	(0.012)	(2.012)
Bank level controls	X	X	X	X	X
Fixed effects	X	X	X	X	X
Year Dummies	X	X	X	X	X
Observations	401	388	388	388	388
R-squared	0.118	0.119	0.109	0.192	0.221

Notes: Robust standard errors in parentheses.

Proportion of political elites on bank board is the total number of political elites divided by total bank board members across all existing boards (board of directors, board of commissioners, audit board, and supervisory board). Post is an indicator equal to one for the post-democratization years (after 1999).

All dependent variables in log form. Lending is measured in billions IDR, return on equity is the ratio of net income to total capital, return on assets is the ratio of net income to total assets, net interest margin is the ratio of net interest income to total assets and bank solvency (as measured by the z-score). Z-score = ((ROA_it+CA)_it/σ(ROA))_i, where CA is the capital adequacy ratio, ROA is averaged for each bank over the pre-and post-democratization periods.

Bank level controls include measures of total board size, the competitiveness of the bank’s industry (a geography-based measure of industry competitiveness using the share of deposits of banks in the industry), size (log (total assets)), operating leverage (fixed assets/total assets), and liquidity (liquid assets/liquid liabilities), and bank age (years from establishment).

*** p < 0.01;

** p < 0.05;

* p < 0.1.

Table 7. The relationship between bank political connections and outcomes for privately held Indonesian banks 1993–2008.

	(1)	(2)	(3)	(4)	(5)
Dependent variable	Lending	Return on equity	Return on assets	Net interest margin	Solvency
Proportion of political elites on bank board * Post	0.045**	0.011*	0.012*	0.014**	0.030**
	(0.002)	(0.003)	(0.006)	(0.001)	(0.018)
Proportion of political elites on bank board	0.078**	0.008**	0.007*	0.009**	0.003*
	(0.003)	(0.001)	(0.002)	(0.002)	(0.001)
Post	−0.112***	−0.148*	−0.111*	−0.121**	−0.052*
	(0.085)	(0.056)	(0.004)	(0.018)	(0.019)
Constant	12.182**	2.889*	0.893*	2.875*	3.958*
	(4.365)	(1.016)	(0.127)	(1.014)	(1.001)
Bank level controls	X	X	.X	X	X
Fixed effects	X	X	X	X	X
Year Dummies	X	X	X	X	X
Observations	1566	1514	1514	1514	1514
R^2	0.196	0.158	0.149	0.241	0.232

Notes: Robust standard errors in parentheses.

Proportion of political elites on bank board is the total number of political elites divided by total bank board members across all existing boards (board of directors, board of commissioners, audit board, and supervisory board). Post is an indicator equal to one for the post-democratization years (after 1999).

All dependent variables in log form. Lending is measured in billions IDR, return on equity is the ratio of net income to total capital, return on assets is the ratio of net income to total assets, net interest margin is the ratio of net interest income to total assets and bank solvency (as measured by the z-score). Z-score = ((ROA_it+CA)_it/σ(ROA))_i, where CA is the capital adequacy ratio, ROA is averaged for each bank over the pre-and post-democratization periods.

Bank level controls include measures of total board size, the competitiveness of the bank’s industry (a geography-based measure of industry competitiveness using the share of deposits of banks in the industry), size (log (total assets)), operating leverage (fixed assets/total assets), liquidity (liquid assets/liquid liabilities), and age (years from establishment).

*** p < 0.01;

** p < 0.05;

* p < 0.1.

Table 7 indicates the results for the privately held banks, including joint venture and private domestic banks.⁷ We do not include foreign-owned banks, as they did not have political connections over the sample period. We find that the relationship between depoliticization and bank performance for privately held banks over this period exhibits the opposite direction as compared to government-owned banks. Namely, political connectedness positively influences lending and performance metrics over this period. Like government-owned banks, the association between political connectedness and bank financial outcomes decreases over time. In other words, the depoliticization of the banking sector decreased the positive influence of political connectedness on performance for privately held banks. While political connectedness is related to more lending and higher performance for privately held banks, this relationship decreases over time. Mirroring government-owned banks, the magnitude of these relationships varies. For example, using the average changes in political connectedness for privately held banks, we find that depoliticization is associated with about 5% decrease in net interest margin and a 10% decrease as a result of this change for return on equity.

Altogether, the data and analysis provide a story of how the depoliticization of the banking sector over democratization period differentially affects banks. Here, we show that the period of democratization was concurrent with a significant decrease in political connectedness, measured both as the number of political connections, and the proportion of board members that are members of the political elite, and that this depoliticization is not primarily due to bank closure.

More importantly, we find that the influence of political connections varies dramatically depending on bank ownership. For government-owned banks, political connectedness negatively impacts bank financial outcomes, while for privately held banks connectedness enhances performance. Over the democratization, however, as political connectedness diminishes, the relationship between political connectedness and financial outcomes diminishes for both sets of banks. Thus, for the government-owned banks, the decrease in political connectedness is beneficial, as the reductive influence of political connectedness diminishes. For privately held banks, however, the positive relationship between political connections and financial outcomes decreases as well. Thus, for these banks, depoliticization is associated with a reduction in this performance-enhancing relationship.

These results provide several insights. First, they suggest that while the literature attempts to describe political connectedness as either negative or positive in terms of relationship to the financial outcomes of banks, the true relationship may vary based on ownership structure; political connections may play a very different role for government-owned banks relative to their privately held counterparts. Thus, evidence here points to the predominance of the political objectives view of political connections for government-owned banks whereby political connectedness represents an avenue through which political entrepreneurs pursue their potentially non-profit-enhancing political ambitions. On the other hand, the positive influence and subsequent decrease in performance of privately held banks as political ties weaken provides evidence of the preferential access view, whereby politically connections serve as profit-enhancing assets for firms.

5. Robustness

5.1. Democratization and financial crisis

The main narrative here is that an intention of democratization is to reduce elite capture of politically important sectors such as banking, and our data and results work to find quantitative evidence of this idea. One main area for discussion is how the financial crisis affects this narrative. Namely, perhaps it is not democratization at all that is related to changes in political connections, but that financial outcomes resulting from crisis are responsible.

First, we would like to acknowledge that it is almost impossible to completely disentangle the financial crisis from democratization as these events/shifts happened one after the other and we do not purport to be able to do this. We do believe, however, that the significant drop in political connections must be related to the change in political structure and we do not believe this to be outlandish. Further, elements of the spirit of democratization that supported better governance were integrated into financial crisis policy. So fully attributing changes to financial crisis policy ignores that democratizing finance was an integral part of financial crisis policy. At least from a generalizability standpoint, however, since financial crisis and democratization often occur in tandem (Gasiorowski 1995), our results are neither unique to this particular circumstance, nor could another analysis necessarily more cleanly identify the effects of democratization separate from financial crisis policy.

We can, however, discuss whether the decrease in political connectedness was the result of financial issues rather than to mitigate political influence in banks. In this sense, financial crisis may have led to depoliticization because political elite on bank boards weakened performance. Perhaps politically connected board members were removed because banks were struggling financially due to the crisis and these board members contributed to poor financial performance. If this were the case, however, then privately held banks would not decrease in their political connectedness because these connections were actually associated with higher performance. Thus it is unlikely that on average the board members were removed purely do to their impact on financial performance.

5.2. Other characteristics of the political elite

It may also be the case that other characteristics of the political elite may be driving their removal from bank boards rather than political connectedness. First, it could be the case that simply the restructuring of the boards is related to change in financial outcomes and not the removal of political elites per se. This would suggest, however, that the particular set of elites in our sample have no special attributes that relate to financial outcomes. We also find this alternative to be unlikely. If this were the case, then we could randomly assign political elite status to those leaving, continuing, and coming onto boards post-democratization, run our analysis again, and it should yield the same results as we have with the political elites. We conduct this robustness check and find that none of the parameters of interest are significant.⁸ Thus, it is something common among this particular group of people that is leading to the results we observe earlier.

If there exists a commonality among the political elites in the sample outside of their connection to the government, however, and this may be related to bank financial outcomes, then the main analysis connecting changes in political ties to the banking sector may also be incorrectly attributed to political connectedness. Perhaps most likely of these is that the political elites are the more educated group of available board members, and their education helps determine bank financial outcomes. While we do not have the CVs of the non-elites to compare our set to, in order for this hypothesis to be valid it must be the case that the variation in education among the political elites would be divided precisely between those on government-owned bank boards and those on privately held boards.

We believe the situation of the first point is rather doubtful; this would mean that the education of elites is related to bank financial outcomes differently for government-owned bank versus privately held banks and precisely in the manner observed in the results. If one assumes education is positively correlated with bank financial outcomes, then the elites on government-owned bank boards would be less educated, and the elites on privately held bank boards would be more educated, as the presence of the former increases performance while the latter decreases. Table 8 shows a t-test in the difference in the mean proportion of board members for government-owned versus privately held over the entire sample period.⁹ While the proportion of elites of privately held banks with advanced education is larger, the difference in these means is not significant.

Table 8. Differences in the proportion of elites on government-owned versus privately held banks with college or postgraduate education.

	(1)	(2)
	Government-owned banks	Privately held banks
College educated only	0.39	0.41
Postgraduate educated	0.28	0.32

Notes: This table reports the mean proportion of college educated and postgraduate educated political elites on government-owned (column 1) and privately held (column 2) bank boards.

* t-test of the differences in the means presented in columns 2 and 3 is significant.

Another characteristic of elites that may be related to outcomes is risk aversion. Similar to the education example, it may be the case that political elites are more (less) risk averse. This possibility is harder to dismiss than education for several reasons. First, risk aversion is a difficult measure, so we cannot determine if there is variability in risk aversion among elites in order to debunk this hypothesis in the same way we did education. Even if we had a metric of risk aversion, it is difficult to completely disentangle the choice of board members who are less risk averse from political connectedness, as the elites may be less risk averse precisely because of their political connections. But again, if risk aversion is actually at the root of the changes in bank financial outcomes observed, it must be the case that those on the government-owned bank board are more or less risk averse while those on the privately held boards are precisely the opposite, which we consider to be unlikely.

5.3. Multiple board memberships

It is worth mentioning the possibility of joint board memberships. If political elites were on multiple boards, then this would influence the way in which we enumerate political connectedness in the data description. However, we found very few of these instances, so we do not consider this to be an issue.

5.4. Other sensitivity tests

We conduct the main analysis for private banks excluding joint venture banks. Inclusion of joint venture banks may bias our results as partial foreign ownership may be influencing performance outcomes rather than the political connections per se. Table 9 shows that removing joint venture banks does not significantly change the results.

Table 9. The relationship between bank political connections and outcomes for privately held Indonesian banks 1993–2008 without joint venture.

	(1)	(2)	(3)	(4)
Dependent variable	Lending	Return on equity	Return on assets	Net interest margin
Proportion of political elites on bank board * Post	0.031**	0.011*	0.013*	0.013**
	(0.001)	(0.003)	(0.005)	(0.004)
Proportion of political elites on bank board	0.075**	0.007*	0.005*	0.009**
	(0.001)	(0.003)	(0.001)	(0.002)
Post	−0.114***	−0.148*	−0.108*	−0.181*
	(0.019)	(0.061)	(0.002)	(0.018)
Constant	12.501**	2.998*	0.983*	3.113*
	(3.215)	(1.007)	(0.224)	(1.002)
Bank level controls	X	X	.X	X
Fixed effects	X	X	X	X
Year Dummies	X	X	X	X
Observations	1124	1101	1101	1101
R^2	0.183	0.153	0.130	0.198

Notes: Robust standard errors in parentheses.

Proportion of political elites on bank board is the total number of political elites divided by total bank board members across all existing boards (board of directors, board of commissioners, audit board, and supervisory board). Post is an indicator equal to one for the post-democratization years (after 1999).

All dependent variables in log form. Lending is measured in billions IDR, return on equity is the ratio of net income to total capital, return on assets is the ratio of net income to total assets, net interest margin is the ratio of net interest income to total assets and bank solvency (as measured by the z-score). Z-score = ((ROA_it+CA)_it/σ(ROA))_i, where CA is the capital adequacy ratio, ROA is averaged for each bank over the pre-and post-democratization periods.

Bank level controls include measures of total board size, the competitiveness of the bank’s industry (a geography-based measure of industry competitiveness using the share of deposits of banks in the industry), size (log (total assets)), operating leverage (fixed assets/total assets), and liquidity (liquid assets/liquid liabilities).

*** p < 0.01;

** p < 0.05;

* p < 0.1.

Finally, it could argued that the transition to democracy actually occurred during the first direct presidential election in 2004 as opposed to 1999. To test this, I consider 2005 the first post-democratization year, and estimate the main equation again. Results are no longer significant using these alternative time periods, suggesting the relationship between political connectedness and bank financial outcomes was significant when comparing pre-and post-democratization years from 1999.¹⁰

6. Conclusion

This analysis sets out to understand how banks respond differentially to the depoliticization of the banking sector over democratization. Essentially, we first establish that this depoliticization occurred by examining the changes over the democratization of the political connectedness of banks and the banking sector, where political connectedness is measured both as the number of unique bank-elite connections as well as changes to the proportion of board members of banks that are elites. Next, we look at how the decrease in political connectedness is related to changes in bank financial outcomes. This is important, as it suggests not only that the depoliticization happened, but also that this change is economically significant, particularly as it applies to banks with different ownership structures.

The results we find support the idea that the democratization period was concurrent with a significant depoliticization of the banking sector using both measures of connectedness. Most importantly, our analysis reveals that political connectedness had a different relationship to bank financial outcomes for the two separate types of banks. Namely, we find that both the performance-reducing and performance-enhancing influence of political connectedness on financial outcomes of government-owned and privately held banks (respectively) diminish over democratization. These results are robust to a variety of robustness checks and sensitivity tests.

This analysis provides several important implications. First, policy-makers would be well advised when crafting regulatory policy during changes in political regime to take pause when implementing a uniform approach for the entire banking sector, as political ties may not be as one-dimensional as previous work shows. Second, this works provides evidence for policy-makers seeking a means to enhance the performance of government-owned banks that dismantling the influence of politicians may be one such avenue. Second, by providing evidence that there exists variation over democratization in the relationship between political connectedness and financial outcomes by bank ownership type, our analysis helps reconcile the main two views in the literature regarding the value of political connections. Namely, we see evidence of the profit-reducing political objectives view for government-owned firms, while privately held firms’ results conform to the preferential access view. Thus, by examining the changes over democratization, we are able to provide an important nuance to this well-discussed relationship.

Perhaps the most pressing limitation of our study is generalizability. On the one hand, the sequence of events in this analysis is only within one context. On the other hand, as mentioned earlier, the transition to democracy exhibits several key similarities over countries and time. Namely, the transition is often fueled by discontentment with elite capture that becomes more apparent as a county’s economy disintegrates during a financial crisis. Thus, while this paper is within one context, parallels to other financial crises followed by regime change can be made.

While this paper provides several important policy relevant insights, other questions remain. We remain agnostic, for instance, on the optimal approach to dismantling political connections. Some of these board members may have left boards voluntarily post-crisis. Others may have been removed from the board by the bank itself. And still others may have been removed from banks by the restructuring from the IBRA post-financial crisis. While it is beyond this scope of this paper to identify optimal methods, understanding whether the approach to changing political connections matter remains a central question in order to understand how to successfully loosen the grip of the political elite in banking over democratization.

Disclosure statement

No potential conflict of interest was reported by the author.

Notes

1. Technological restrictions necessitated the hand-entry of this online information.

2. The most pressing issue in this process is that many Indonesians are called by one name only, which makes matching less precise. In these cases, the position at the bank was verified by directly examining the CV itself.

3. We cannot replicate the statistics in Table 4, as closed bank have no post-democratization proportion by definition.

4. This methodology was favored over triple interaction effects for ease of interpretation.

6. Calculation available by request from author.

7. In the robustness section, we remove joint venture banks and conduct the analysis again. Results remain robust to this sensitivity test.

8. Results excluded here for brevity are available by request from the author.

9. A proportion is used to control for board size.

10. Results here excluded for brevity.

11. Translated from Bahasa Indonesia by a research assistant, Edwin Thong, after interviewing the Vice President of Tokoh Indonesia via phone.

Appendix 1. Details on data sources

The primary source of data is the Encyclopedia of Prominent Indonesians [Encyclopedia]. The Encyclopedia is a comprehensive online database of profiles of the most Indonesian contemporary and historical political figures. The database was created and is maintained by a non-profit media consulting firm whose aim it is to increase political transparency in contemporary Indonesia. Politicians and other elites are included only if primary source information and validation is obtained via direct interviews with the politicians or a family member of the politician. A summary of the method employed to add a politician is the following: Persons of political influence to be included are suggested and discussed among the consulting firm’s executive officers and staff members, all of whom are current or former journalists specializing in politics. The idea is to include those who have held significant political offices as well as those who may not have directly have been an elected politician, but yield considerable political influence as important members of the military, private sector, a political party, or are significant bureaucrats. Then, the person is contacted for a face-to-face interview. If the person cannot meet in person, the interview is conducted over the phone. If that is not possible, their direct family (spouse, parents or siblings) is contacted and an interview is requested. Persons are only included if either of these primary sources are available. Further details on the interview process can be found in Appendix 1.

To check internal consistency and add to the Encyclopedia, an additional data-set of CVs of the most influential politicians was obtained privately from a prominent political risk consulting firm in Indonesia. Finally, an additional confidential data-set of CVs of the members of Indonesian bicameral legislature was obtained from the Indonesian government, as newly elected parliamentarians are asked to submit a CV once elected. The information from the Encyclopedia was hand-entered in 2009 and updated annually until 2012.

The Encyclopedia of Prominent Indonesians was created and is maintained by Tokoh Indonesia, a media consulting firm with the mission to “create greater transparency”¹¹ of the political sector in Indonesia. Its members and staff are largely journalists specializing in politics. The Encyclopedia of Prominent Indonesians includes information about the most influential formal and informal leaders (both current and historic) involved in politics in Indonesia as determined by Tokoh’s staff. Each profile includes the individual’s curriculum vitae and other personal information. Tokoh Indonesia describes the standard procedure for adding a profile to the Encyclopedia as follows. Based on the editor-in-chief’s and the staff’s collective expertise, a politician is considered for addition to the Encyclopedia. There is no limit to the number of persons who could in principle be included in their search. After a politician is considered worth adding, Tokoh Indonesia collects primary source data on the CVs of politicians by interviewing the person of interest in person or if unavailable in person, by phone. If the person cannot be reached, Tokoh Indonesia then interviews his or her family members also in person first and if unavailable then by phone. If primary sources are entirely unavailable, two main secondary sources are used. First, the Indonesian Ministry of Social Affairs may provide Tokoh Indonesia with politicians CVs by request. Second, Tokoh searches the following eight newspapers for information regarding the politicians CV: Kompas, Republika, Suara Pembaruan, Media Indonesia, Indopos, Tempo, Gatra, and Berita Indonesia. Tokoh also searches the secondary sources for information for those politicians whom primary sources were available for any additional information or discrepancies. Any additional or conflicting information is confirmed via phone or text with the respective individuals or their families by sending them document copies of the information. If the secondary information cannot be verified or discrepancies not clarified by primary sources, the additional information is not included in the Encyclopedia. Tokoh Indonesia maintains open communication with the profiled individuals and their families in case they need to correct or add any information on the website.

The information from each profile in the Encyclopedia was hand-entered. The bulk of the data was entered in 2009, and each year it was update to include politicians added to the Encyclopedia each year until 2012. Data included information regarding the politician’s personal, educational, private sector, and political background. Personal information includes year of birth, birthplace, name of spouse and children, and religious affiliation. Educational background includes the dates, location, and institutions for each degree obtained. Private sector/non-government background includes information regarding the title of the position, the name of the company or organization, and the dates for each non-governmental position held.

Next, the information from the main data-set was verified using additional data provided by an independent company conducting political risk analysis in Indonesia, PT Reformasi Info Sastra [PT Ris]. One of the leading political risk consulting firms in Indonesia, PT Ris specializes in the analysis of investment conditions, providing clients with strategic consulting, customized research, and syndicated reports. The data used are from Pt Ris’s book Who’s Who in the Yudhoyono Era [Who’s Who], which provides the CVs of more than 140 government officials, policy-makers, and politicians, including the entire cabinet, security officials, the leaders of major state institutions, senior civil servants, political party chairs, parliamentary faction heads, and major state enterprise directors. Almost all of the politicians from Who’s Who appear in the Tokoh Indonesia data-set. Discrepancies in information between the two data-sets were minor.

The third source of data is the CVs of all members of the bicameral Indonesia legislature (the House of Representatives (DPR) and House of Regional Representatives (DPD), obtained confidentially from a source in the Indonesian Government. All members of the DPR and DPD are asked to provide a CV to the government once elected to office. Again, this was used to both verify and supplant the Encyclopedia.